Page 176 - Read Online

Research Article | Open Access

Journal of Materials Informatics

Hei et al. J. Mater. Inf. 2026, 6, 15 DOI:10.20517/jmi.2025.75

Enhanced multi-tuple extraction for materials:
integrating pointer networks and augmented
attention

Mengzhe Hei , Zhouran Zhang 2,#,* , Qingbao Liu , Yan Pan , Xiang Zhao , Yongqian Peng , Yicong Ye , Xin
1,#
2
1
3
2
3
Zhang , Shuxin Bai 2
1,*
Keywords:
AI for materials, multi-tuple
extraction, MatSciBERT,
attention mechanism
Citation: Hei, M.; Zhang, Z.;
Liu, Q.; Pan, Y.; Zhao, X.;
Peng, Y.; Ye, Y.; Zhang, X.;
Bai, S. Enhanced multi-tuple
extraction for materials:
integrating pointer networks
and augmented attention. J.
Mater. Inf. 2026, 6, 15.
https://dx.doi.org/10.20517
/jmi.2025.75

Received: 28 Aug 2025
Accepted: 30 Sep 2025
Published: 30 Mar 2026
Abstract
Academic Editors: Extracting reliable, tuple-level information from materials texts is essential for data-driven
Sheng Sun, Hao Li materials design, yet multi-tuple sentences remain difficult due to intertwined semantics,
Copy Editor: syntactic complexity, and sparse supervision in higher-density cases. In this study, we
Pei-Yun Wang
Production Editor: address these challenges by formulating information extraction as an integrated process
Pei-Yun Wang that couples entity extraction with tuple allocation. The framework combines an entity
extraction module based on bidirectional encoder representations from transformers
(MatSciBERT) with pointer networks and an allocation module that models inter- and
intra-entity attention to enforce tuple coherence. Using the mechanical properties of
multi-principal element alloys as a case study, we define the target schema and evaluate
exact match tuple accuracy. Our experiments demonstrate F1 scores of 0.96, 0.95, 0.85,
and 0.75 on datasets containing one to four tuples per sentence, and 0.85 on a randomly
curated set. Ablation studies show that the allocation module is most critical, with
inter-entity attention contributing more than intra-entity attention. Error analysis attributes

1 National Key Laboratory of Information Systems Engineering, National University of Defense Technology, Changsha 410072, Hunan,
China.
2 College of Aerospace Science and Engineering, National University of Defense Technology, Changsha 410072, Hunan, China.
3 Laboratory for Big Data and Decision, National University of Defense Technology, Changsha 410072, Hunan, China.
# These authors contributed equally to this work.

* Correspondence to: Assoc. Prof. Zhouran Zhang, College of Aerospace Science and Engineering, National University of Defense
Technology, Changsha 410072, China. E-mail: zzhang_nudt@outlook.com; Prof. Xin Zhang, National Key Laboratory of Information
Systems Engineering, National University of Defense Technology, Changsha 410072, Hunan, China. E-mail: shinezhang_nudt@163.com

www.oaepublish.com Submit a Manuscript: https://ucenter.oaepublish.com